1 Introduction

1.1 The Information Integration Challenge

Information integration is a challenge that affects many organizations. Many run several different database systems. Each of these systems stores data and has a set of applications that run against it. This data is just bits and bytes on a file system - and only a database can turn the bits and bytes of data into business information. Integration and consolidation of all business information would allow an organization to easily and quickly take advantage of the synergies inherent in business information.

Consolidation of all data into one database system is often difficult. This is in large part because many of the applications that run against one database may not have an equivalent that runs against another. Until such time as migration to one consolidated database system is made feasible, it is necessary for the various heterogeneous database systems to work together.

There are several problems to overcome before such interoperability becomes possible. The database systems can have different access interfaces, different data types, different capabilities, and different ways of handling error conditions. Even when one relational database is trying to access another relational database the differences are significant. In such a situation, the common features of the databases include data access through SQL, two-phase commit, and similar data types.

However, there are significant differences as well. SQL dialects can be different, as can transaction semantics. There can be some data types in one database that do not exist in the other. The most significant area of difference is in the data dictionaries of the two databases. Most data dictionaries contain similar information but the information is structured for each in a completely different way. There are several possible ways of overcoming this problem. In this book, we describe the approach that Oracle has taken for synchronously accessing information from multiple sources.

1.2 How Oracle Addresses Synchronous Information Integration

If a client program needs to access or modify data at several Oracle databases, it can open connections to each of them. This approach, however, has several drawbacks; among them are the following:

• If you want to join data from the databases, the client must contain logic allowing that.

• If data integrity must be guaranteed, the client needs to contain transaction coordination logic.

Oracle provides another approach called distributed processing, where the client connects to one Oracle database and shifts the burden of joining data and transaction coordination to that database. The database to which the client program connects is called the local database. Any database other than this one is a remote database. The client program can access objects at any of the remote databases using database links. The Oracle query processor takes care of the joins and its transaction engine takes care of the transaction coordination.

The approach that Oracle has taken to solving the heterogeneous connectivity problem is to allow a non-Oracle system to be one of the remote nodes in the previously described scenario. From the client's point of view, the remote non-Oracle system functions like a remote Oracle system. It appears to understand the same SQL dialect and to have the same data dictionary structure as an Oracle system. Access to a non-Oracle system in this manner is done through a component in the Oracle server called Heterogeneous Services.

The work done by the Heterogeneous Services component is, for the most part, completely transparent to the end user. With only a few exceptions (these are noted in later chapters), you are not required to do anything different to access a non-Oracle system than is required for accessing an Oracle system. The Heterogeneous Services component is used as the foundation for implementing Oracle's access to non-Oracle databases.

An Oracle Database Gateway works in conjunction with the Heterogeneous Services component of the Oracle Database server to access a particular, commercially available, non-Oracle system for which that Oracle Database Gateway has been designed. For example, you use the Oracle Database Gateway for Sybase to access a Sybase database. Oracle also provides an Oracle Database Gateway for ODBC which enables you to use ODBC drivers to access non-Oracle databases.

Using an Oracle Database Gateway, you can access data anywhere in a distributed database system without being required to know either the location of the data, or how it is stored.

Oracle also offers asynchronous information integration solutions that are mentioned here, but that are not discussed in this book. Briefly, these solutions include:

• Oracle Streams

  Oracle Streams enables the propagation of data, transactions, and events in a single data stream or queue, either within a database or between multiple databases. Not only can Oracle Streams capture, propagate, and apply changes to data, it can also handle data structure changes (DDL) and user-defined events. Changes can be captured and applied as is, or transformed at any point in the capture, propagation, and application processing.

• Messaging Gateway

  The messaging gateway enables communication between Oracle and other non-Oracle message queuing.

• Open System Interfaces

  Oracle offers a number of open interfaces, such as OCI, JDBC, and ODBC, that enable customers to use third party applications or to write their own client applications to access the Oracle database.

1.3 Benefits of Oracle's Solution for Synchronous Information Integration

Much of the processing power of Oracle Database Gateways is integrated into the database. This provides an efficient solution for information integration that enables full exploitation of the power and features of the Oracle database. This includes such features as powerful SQL parsing and distributed optimization capabilities.

The following sections explore the benefits of Oracle's approach to resolving the challenges of a heterogeneous environment:

• Remote Data Can Be Accessed Transparently

• No Unnecessary Data Duplication

• SQL Statements Can Query Several Different Databases

• Oracle's Application Development and End User Tools Can Be Used

• Users Can Talk to a Remote Database in its Own Language